DSXchange

If you google the UNIX "sed" command, you'll see that changing the suggest command to should do the trick.

Is this an Oracle VarChar2() column and have you looked into the $APT_STRING_PADCHAR setting?

replace the command with the more efficient

The CFF stage is set up differently, but you only need to use it if you have OCCURS DEPENDING ON or variable REDEFINES; otherwise just use the sequential file stage and set the default value for the columns in question.

The error regarding the previously undefined variable is an old one (I'm really surprised it isn't fixed yet), but the abort within the DSD.WriteLog shouldn't be happening. If the mount point for your project directory is not close to full then it might be a corrupt log file, the easiest way to fix ...

And since you are in Server, you can read the DB2 table as suggested above, then use a named-pipe sequential file for the cff source and thus don't even need to land your data on disk.

The easiest way is to attach to your project directory, then do a 'find *{Transform name}*'

Probably some of your jobs aren't using or getting passed the $APT_CONFIG_FILE parameter.

As Ray has stated, once you go to NLS you can't go back. Jobs exported from a NLS project need to be imported into a NLS project.

The '-' sign in your output line was confusing, it made it look like negative value lists were being returned.

I can't check now, but perhaps the pivot stage is incompatible. What happens if you write a small dummy job using the px pivot and seeing if you get the same abort?

There should be other messages in the log file which hint at the cause of the problem, the message you posted is just documenting an effect. How heavily loaded is your system, are you getting timeouts in DataStage?

Approaching the problems one-by-one. At Version 8.1 you can store your metadata in xmeta instead of in DataStage hashed files, I haven't experimente with this but I would guess that in that case no dynamic files would be open, perhaps just the static VOC hashed file. Unless you see this problem prio...

This is a matter of using the 'default' attribute for each column.

The path to the source code is listed in the compile messages, could you post that to this thread?

(there is no need to open up a new thread, just reply to the previous one)

Again, could you post an example. Is the data in the form of:

Acct1,Date1
Acct1,Date2
...
Acct2,Date1

or

Acct1,Date1,Date2,Date3,...

or

"Acct1","Date1,Date2,Date3"